Vaccines are urgently needed to slow the spread of the HIV/AIDS epidemic, and immunogens that can induce neutralizing antibodies remain an important goal of vaccine research. An increasing number of human monoclonal antibodies (mAbs) have been isolated in recent years from individuals who show remarkably broad and potent neutralizing responses. Coupled with past work, these results support the idea that the human immune system can generate rare but potent protective antibodies to the virus. Much work, including our own, has focused on identifying forms of the HIV-1 envelope glycoprotein (Env) that bind to broadly neutralizing antibodies. In contrast to approaches that optimize Env for binding to these highly affinity-matured antibodies, we and others now propose to identify immunogens that can bind to the germline form of the broadly neutralizing antibodies as a more productive approach to immunization. While three-dimensional structures of antibody-antigen interactions have informed antigen design, the corresponding germline antibody sequences rarely show any binding to the HIV-1 Env. As a result, structural information is not readily available for rational design of germline immunogens. We propose therefore to employ a directed molecular evolution approach to the problem of identifying germline-specific immunogens. Such immunogens could stimulate immature B cells that carry the germline precursor of a mature broadly neutralizing antibody. Because any individual germline sequence represents only a few percent of all rearranged immunoglobulin genes, activating specific B cells with germline-specific immunogens could increase the likelihood of inducing an appropriate antibody. One can also envision using a series of immunogens that stimulate partially affinity-matured antibodies along a particular path to form a mature neutralizing antibody. The recently characterized VRC01 mAb is well suited to this approach, notably because the CDRH3 domain is less important for the neutralization activity compared to most of the other broadly neutralizing mAbs. This minimizes one of the constraints in predicting the structure of the germline VRC01 precursor. We will create a number of VRC01-like mAbs that are increasing reverted to the germline sequence and use these as reagents to screen libraries of Env variants created by in vitro homologous DNA recombination. The revertants can be used in a stepwise recursive fashion, whereby Env variants that bind minimally reverted mAbs can be used as parents to create libraries of additional variants for screening with more highly reverted forms. We will evaluate the ability of these germline-specific Env immunogens to bind to myeloma cells that express VRC01-like germline antibodies on the cell surface, an interaction that resembles one of the first steps of B-cell activation. This Proposal represents an innovative approach to the identification of candidate immunogens for HIV-1 vaccines. It has general applicability to other pathogens for which protective (or broadly protective) mAbs have been identified (e.g., RSV or influenza) but induction of those antibody specificities has proven problematic.